planning

Planning - Implementation Planning and Step Breakdown

Prerequisite Protocol

Mandatory Trigger Rules

Trigger Timing: After prompt-enhancer completes requirement refinement (mandatory)

Trigger Conditions：

✅ Clear requirements and solution already exist
✅ Implementation planning and step breakdown are needed
✅ Complex task (requires multiple steps to complete)

Exceptions：

Simple task (can be completed in a single step)
User explicitly states "start directly"

Dependency Check

Prerequisite Dependencies：prompt-enhancer

Check Content：

Check artifact file:memory/artifacts/prompt-enhancer/
Confirm that clear requirements and a selected solution already exist
If there is no prerequisite artifact, prompt-enhancer must be executed first

Collaboration with goal-oriented

Trigger Source: Mandatory invocation by goal-oriented skill

Trigger Flow:

goal-oriented 创建目标
    ↓
调用 prompt-enhancer（需求细化）
    ↓
生成需求方案工件
    ↓
调用 planning（实施规划）
    ↓
读取需求方案工件 → 生成实施计划 → plan-review
    ↓
生成实施计划工件
    ↓
传递给 execution 阶段
    ↓
任务完成后 → experience-manager 经验沉淀

Artifact Reception:

goal-oriented will write the artifact to memory/artifacts/prompt-enhancer/
planning automatically detects and reads the latest artifact

Artifact Transfer Mechanism

Input Artifact (from prompt-enhancer)

Artifact Path: memory/artifacts/prompt-enhancer/result-{timestamp}.json

Artifact Format:

{
  "skill": "prompt-enhancer",
  "version": "2.0.0",
  "requirement": {
    "original": "用户原始需求",
    "refined": {
      "type": "功能开发|重构|优化",
      "core_features": ["功能1", "功能2"],
      "target_users": "目标用户",
      "platform": "平台",
      "tech_stack": ["技术1", "技术2"],
      "success_criteria": ["标准1", "标准2"],
      "timeline": "时间线"
    }
  },
  "solution": {
    "selected": "选定方案",
    "reasons": ["理由1", "理由2"],
    "alternatives": [
      {
        "name": "备选方案1",
        "pros": ["优势"],
        "cons": ["劣势"]
      }
    ]
  }
}

Read Method:

# 自动检测最新的 prompt-enhancer 工件
PROMPT_ENHANCER_ARTIFACT=$(ls -t memory/artifacts/prompt-enhancer/*.json | grep -v latest.json | head -1)

if [ -n "$PROMPT_ENHANCER_ARTIFACT" ]; then
  echo "✅ Found prompt-enhancer artifact: $PROMPT_ENHANCER_ARTIFACT"
  # Read artifact content using Read tool
else
  echo "❌ No prompt-enhancer artifact found"
  echo "⚠️ Must execute prompt-enhancer before planning"
fi

Output Artifact (passed to subsequent skills)

Artifact Path: memory/artifacts/planning/result-{timestamp}.json

Artifact Format: See "Step 8: Generate Artifact"

Transfer Mechanism:

Create artifact file
Create latest.json symbolic link
Subsequent skills can automatically detect latest.json

Overview

planning is a mandatory prerequisite skill, responsible forimplementation planningandstep breakdown. It transforms clear requirements and solutions into executable implementation plans, ensures plan feasibility through plan-review, and reduces execution risk.

Core Capabilities：

Step Breakdown: Break down requirements into executable steps
Dependency Identification: Identify dependencies between steps
Resource Planning: Clarify required resources and time
Risk Assessment: Identify potential risks and countermeasures
plan-review: User confirms the implementation plan

Core Principle: Plan before acting; planning comes first. A good plan is half the success.

Key Value：

Clarify the execution path and avoid blind action
Identify dependencies and arrange the order reasonably
Assess risks in advance and prepare countermeasures
User confirms the plan to ensure consistent understanding

When to Use

Mandatory Usage Scenarios：

After prompt-enhancer is completed
Requirements are clear and the solution has been selected
The task requires multiple steps to complete

Non-Usage Scenarios：

Simple task (can be completed in a single step)
User explicitly states "start directly"
A detailed implementation plan already exists

The Process

digraph planning_process {
    rankdir=TB;

    "接收需求方案工件" [shape=box, style=filled, fillcolor="#c8e6c9"];
    "步骤分解" [shape=box, style=filled, fillcolor="#bbdefb"];
    "依赖识别" [shape=box, style=filled, fillcolor="#fff9c4"];
    "资源规划" [shape=box, style=filled, fillcolor="#ff9800"];
    "风险评估" [shape=box, style=filled, fillcolor="#ff9800"];
    "生成实施计划" [shape=box, style=filled, fillcolor="#f8bbd0"];
    "plan-review" [shape=diamond, style=filled, fillcolor="#e1bee7"];
    "用户确认" [shape=diamond, style=filled, fillcolor="#e1bee7"];
    "调整计划" [shape=box, style=filled, fillcolor="#f8bbd0"];
    "生成工件" [shape=box, style=filled, fillcolor="#81c784"];
    "传递给执行技能" [shape=doublecircle, style=filled, fillcolor="#81c784"];

    "接收需求方案工件" -> "步骤分解";
    "步骤分解" -> "依赖识别";
    "依赖识别" -> "资源规划";
    "资源规划" -> "风险评估";
    "风险评估" -> "生成实施计划";
    "生成实施计划" -> "plan-review";
    "plan-review" -> "用户确认";
    "用户确认" -> "调整计划" [label="需调整"];
    "用户确认" -> "生成工件" [label="确认"];
    "调整计划" -> "生成实施计划";
    "生成工件" -> "传递给执行技能";
}

Step 1: Receive Requirement Solution Artifact

Read the artifact from the prerequisite skill prompt-enhancer:

// memory/artifacts/prompt-enhancer/result-{timestamp}.json
{
  "requirement": {
    "original": "做一个用户认证功能",
    "refined": { ... }
  },
  "solution": {
    "selected": "方案2：微信云开发",
    "reasons": [ ... ]
  }
}

Step 2: Step Breakdown

Break down requirements into executable implementation steps

Breakdown Principles

MECE Principle: Mutually exclusive, collectively exhaustive

Each step is independent and non-overlapping
All steps together cover the complete requirements

Granularity control：

Each step can be completed in 0.5-2 days
Steps are not too fine-grained (avoid over-planning)
Steps are not too coarse (avoid ambiguity)

Decomposition methods

Method 1: Decompose by functional module

用户认证功能
├─ 注册模块
│   ├─ 步骤1：实现手机验证码发送
│   └─ 步骤2：实现注册流程
├─ 登录模块
│   ├─ 步骤3：实现手机验证码登录
│   └─ 步骤4：实现微信OAuth
└─ 找回密码模块
    └─ 步骤5：实现找回密码流程

Method 2: Decompose by chronological order

Week 1: 基础功能
├─ Day 1-2：环境配置 + 验证码发送
├─ Day 3-4：注册登录流程
└─ Day 5：找回密码

Week 2: 优化测试
├─ Day 1-2：测试与Bug修复
├─ Day 3-4：性能优化
└─ Day 5：上线准备

Method 3: Decompose by technical layer

前端
├─ 步骤1：登录UI组件
├─ 步骤2：表单验证
└─ 步骤3：API对接

后端
├─ 步骤4：云函数开发
├─ 步骤5：数据库设计
└─ 步骤6：第三方集成

Step 3: Dependency identification

Identify dependencies between steps

Dependency types

Dependency type	Description	Example
Strong dependency	The prerequisite step must be completed first	Registration → Login
Weak dependency	Best to complete first, but can be done in parallel	UI design ↔ API design
No dependency	Can be executed independently	Documentation writing

Dependency graph example

步骤1：环境配置
    ↓
步骤2：验证码发送
    ↓
┌─────────┴─────────┐
↓                   ↓
步骤3：注册流程    步骤4：登录流程
    ↓                   ↓
    └─────────┬─────────┘
              ↓
        步骤5：找回密码
              ↓
        步骤6：测试上线

Step 4: Resource planning

Clarify the resources required for each step

Resource dimensions

Human resources：

Who will execute it? (Frontend, backend, testing)
How many people are needed?
Is collaboration needed?

Time resources：

How long is each step expected to take?
Overall timeline?
Key time milestones?

Technical resources：

What tech stack is needed?
Are third-party services needed?
Are new tools needed?

Financial resources：

Are there costs? (Cloud services, SMS services)
Budget constraints?

Resource planning table

Step	Personnel	Time	Technology	Cost
Environment setup	1 person	0.5 days	WeChat Developer Tools	None
Verification code sending	1 person	1 day	Alibaba Cloud SMS Service	~¥0.05/message
Registration flow	1 person	2 days	Cloud development, cloud functions	None
Login flow	1 person	2 days	OAuth、JWT	None
Password recovery	1 person	1 day	-	None
Testing and launch	1 person	1.5 days	-	None

Step 5: Risk assessment

Identify potential risks and countermeasures

Risk types

Risk type	Risk description	Probability	Impact	Countermeasures
Technical risk	Complexity of WeChat OAuth integration	Medium	High	Study the documentation in advance, reserve buffer
Time Risk	SMS service review time	High	Medium	Apply in advance and prepare a backup plan
Cost Risk	SMS costs exceed budget	Low	Low	Set sending frequency limits
Integration Risk	Difficulty integrating with existing systems	Medium	High	Prioritize researching existing system interfaces

Risk Matrix

影响 ↑
高  │  [OAuth集成]     [系统对接]
    │      中风险        高风险
中  │  [审核时间]      [成本超支]
    │      中风险        低风险
低  │
    └──────────────────────→ 概率
         低        中        高

Step 6: Generate Implementation Plan

Integrate all content to generate a complete implementation plan

Implementation Plan Template

# 实施计划

## 基本信息
- **项目名称**: 用户认证系统
- **目标**: 实现注册、登录、找回密码功能
- **时间线**: 2周（10个工作日）
- **团队**: 1人（全栈）

---

## 实施步骤

### Phase 1: 环境准备（Day 1）

#### 步骤1：环境配置
**目标**: 搭建开发环境

**任务**:
- [ ] 创建微信小程序项目
- [ ] 配置云开发环境
- [ ] 初始化云数据库
- [ ] 配置项目目录结构

**产出**:
- 可运行的云开发项目框架
- 数据库表结构

**依赖**: 无

**风险**: 无

---

### Phase 2: 核心功能（Day 2-6）

#### 步骤2：验证码发送（Day 2）
**目标**: 实现手机验证码发送功能

**任务**:
- [ ] 申请阿里云短信服务
- [ ] 开发云函数：send-code
- [ ] 前端调用接口
- [ ] 测试验证码发送

**产出**:
- send-code 云函数
- 前端调用封装

**依赖**: 步骤1

**风险**: 短信服务审核（应对：提前申请）

---

#### 步骤3：注册流程（Day 3-4）
**目标**: 实现用户注册功能

**任务**:
- [ ] 设计注册UI（手机号、验证码、密码）
- [ ] 开发云函数：register
- [ ] 表单验证逻辑
- [ ] 数据存储
- [ ] 测试注册流程

**产出**:
- 注册页面
- register 云函数
- 用户数据表

**依赖**: 步骤2

**风险**: 无

---

#### 步骤4：登录流程（Day 5-6）
**目标**: 实现手机验证码和微信OAuth登录

**任务**:
- [ ] 设计登录UI
- [ ] 开发云函数：login（验证码登录）
- [ ] 开发云函数：wechat-login（OAuth）
- [ ] Token生成与存储
- [ ] 测试登录流程

**产出**:
- 登录页面
- login 和 wechat-login 云函数
- Token管理

**依赖**: 步骤3

**风险**: 微信OAuth集成复杂度（应对：预留buffer）

---

### Phase 3: 补充功能（Day 7-8）

#### 步骤5：找回密码（Day 7-8）
**目标**: 实现找回密码功能

**任务**:
- [ ] 设计找回密码UI
- [ ] 开发云函数：reset-password
- [ ] 验证码验证
- [ ] 密码重置逻辑
- [ ] 测试找回密码流程

**产出**:
- 找回密码页面
- reset-password 云函数

**依赖**: 步骤2

**风险**: 无

---

### Phase 4: 测试上线（Day 9-10）

#### 步骤6：测试与优化（Day 9）
**目标**: 全面测试，修复Bug

**任务**:
- [ ] 功能测试（注册、登录、找回密码）
- [ ] 边界测试（验证码过期、错误处理）
- [ ] 性能测试（响应时间、并发）
- [ ] Bug修复

**产出**:
- 测试报告
- Bug修复清单

**依赖**: 步骤5

**风险**: 发现重大Bug（应对：预留1天buffer）

---

#### 步骤7：上线准备（Day 10）
**目标**: 准备上线，文档完善

**任务**:
- [ ] 代码审查
- [ ] 性能优化
- [ ] 文档编写
- [ ] 提交审核

**产出**:
- 完整代码
- 技术文档
- 用户手册

**依赖**: 步骤6

**风险**: 审核不通过（应对：提前了解审核规范）

---

## 依赖关系图

Step 1: Environment Configuration ↓ Step 2: Verification Code Sending ↓ ┌─────────┴─────────┐ ↓ ↓ Step 3: Registration Flow Step 4: Login Flow ↓ ↓ └─────────┬─────────┘ ↓ Step 5: Password Recovery ↓ Step 6: Testing ↓ Step 7: Launch


---

## 资源规划

| 步骤 | 人员 | 时间 | 关键技术 |
|------|------|------|---------|
| 环境配置 | 1人 | 0.5天 | 微信开发者工具 |
| 验证码发送 | 1人 | 1天 | 阿里云短信 |
| 注册流程 | 1人 | 2天 | 云开发 |
| 登录流程 | 1人 | 2天 | OAuth、JWT |
| 找回密码 | 1人 | 1天 | - |
| 测试 | 1人 | 1天 | - |
| 上线 | 1人 | 0.5天 | - |

**总计**: 1人 × 10天

---

## 风险与应对

| 风险 | 概率 | 影响 | 应对措施 |
|------|------|------|---------|
| 短信服务审核 | 高 | 中 | 提前申请，准备备选方案 |
| OAuth集成复杂 | 中 | 高 | 提前研究文档，预留buffer |
| 系统对接困难 | 中 | 高 | 优先调研现有系统 |
| 发现重大Bug | 低 | 高 | 预留1天buffer |

---

## 关键里程碑

- **Day 1**: 环境准备完成
- **Day 4**: 注册功能完成
- **Day 6**: 登录功能完成
- **Day 8**: 所有功能开发完成
- **Day 10**: 测试通过，提交审核

---

## 成功标准

- ✅ 注册、登录、找回密码功能完整
- ✅ 登录响应时间 < 2秒
- ✅ 无P0/P1级Bug
- ✅ 通过微信审核

Step 7: plan-review (mandatory)

plan-review It is a mandatory step to ensure the implementation plan is confirmed by the user

plan-review process

Present the plan：

## 📋 实施计划已完成

### 核心内容
- **总时长**: 10个工作日
- **步骤数**: 7个步骤
- **关键里程碑**: 4个
- **风险点**: 4个

### 关键步骤
1. 环境配置（Day 1）
2. 验证码发送（Day 2）
3. 注册流程（Day 3-4）
4. 登录流程（Day 5-6）
5. 找回密码（Day 7-8）
6. 测试（Day 9）
7. 上线（Day 10）

### 潜在风险
⚠️ 短信服务审核时间
⚠️ 微信OAuth集成复杂度

---

**请确认实施计划：**
- A) ✅ 确认，开始执行
- B) 需要调整（请说明具体问题）
- C) 需要重新规划

After user confirmation：

✅ User confirms → generate artifact, pass to subsequent skills
❌ User requests adjustments → adjust the plan based on feedback
❌ User requests replanning → re-execute planning

plan-review tool call (AskUserQuestion)

Use the AskUserQuestion tool to present the plan and obtain user confirmation:

AskUserQuestion(
  questions=[
    {
      "question": "实施计划已完成，请确认是否开始执行？",
      "header": "计划确认",
      "options": [
        {
          "label": "✅ 确认，开始执行",
          "description": "同意实施计划，生成工件并传递给后续技能"
        },
        {
          "label": "需要调整",
          "description": "对部分步骤有疑问，需要修改计划（请说明具体问题）"
        },
        {
          "label": "重新规划",
          "description": "对整体方案不满意，需要重新分解步骤"
        }
      ],
      "multiSelect": false
    }
  ]
)

User response handling:

User selects "✅ Confirm" → generate artifact, pass to subsequent skills
User selects "Needs adjustment" → modify the plan based on user feedback
User selects "Replan" → re-execute the planning process

Step 8: Generate Artifact

Create the artifact file and pass it to subsequent skills:

{
  "skill": "planning",
  "version": "1.0.0",
  "timestamp": "2026-03-25T10:30:00Z",
  "project": {
    "name": "用户认证系统",
    "goal": "实现注册、登录、找回密码功能",
    "timeline": "2周（10个工作日）",
    "team": "1人（全栈）"
  },
  "implementation_plan": {
    "phases": [
      {
        "name": "Phase 1: 环境准备",
        "steps": [
          {
            "id": 1,
            "name": "环境配置",
            "duration": "0.5天",
            "tasks": ["创建小程序项目", "配置云开发", "初始化数据库"],
            "dependencies": [],
            "risks": []
          }
        ]
      },
      {
        "name": "Phase 2: 核心功能",
        "steps": [
          {
            "id": 2,
            "name": "验证码发送",
            "duration": "1天",
            "tasks": ["申请短信服务", "开发send-code云函数", "前端调用"],
            "dependencies": [1],
            "risks": ["短信服务审核"]
          },
          {
            "id": 3,
            "name": "注册流程",
            "duration": "2天",
            "tasks": ["设计UI", "开发register云函数", "表单验证"],
            "dependencies": [2],
            "risks": []
          }
        ]
      }
    ],
    "dependencies_graph": {
      "1": [],
      "2": [1],
      "3": [2],
      "4": [2, 3],
      "5": [2],
      "6": [5],
      "7": [6]
    },
    "total_steps": 7,
    "total_duration": "10天"
  },
  "resource_plan": {
    "human": {
      "team_size": 1,
      "roles": ["全栈开发"]
    },
    "time": {
      "total_days": 10,
      "start_date": "2026-03-25",
      "end_date": "2026-04-07"
    },
    "technology": [
      "微信小程序",
      "云开发",
      "阿里云短信",
      "OAuth",
      "JWT"
    ],
    "cost": {
      "短信服务": "~¥0.05/条",
      "云开发": "免费额度"
    }
  },
  "risk_assessment": [
    {
      "risk": "短信服务审核时间",
      "probability": "high",
      "impact": "medium",
      "mitigation": "提前申请，准备备选方案"
    },
    {
      "risk": "微信OAuth集成复杂度",
      "probability": "medium",
      "impact": "high",
      "mitigation": "提前研究文档，预留buffer"
    }
  ],
  "milestones": [
    {"day": 1, "name": "环境准备完成"},
    {"day": 4, "name": "注册功能完成"},
    {"day": 6, "name": "登录功能完成"},
    {"day": 8, "name": "所有功能开发完成"},
    {"day": 10, "name": "测试通过，提交审核"}
  ],
  "review_status": "confirmed",
  "next_skills": ["experience-manager"]
}

Planning Methodology

WBS Work Breakdown Structure

Work Breakdown Structure：

项目
├─ 阶段1
│   ├─ 任务1.1
│   ├─ 任务1.2
│   └─ 任务1.3
├─ 阶段2
│   ├─ 任务2.1
│   └─ 任务2.2
└─ 阶段3
    ├─ 任务3.1
    └─ 任务3.2

Critical Path Method (CPM)

Identify the critical path that affects the total duration:

环境配置(0.5天) → 验证码(1天) → 注册(2天) → 登录(2天) → 测试(1天) → 上线(0.5天)
                     ↓
                 找回密码(1天)

Critical Path: Environment → Verification Code → Registration → Login → Testing → Launch = 7 days

Risk Management Framework

Identify → Assess → Respond → Monitor

Identify: List all potential risks
Assess: Probability × Impact
Respond: Avoid, transfer, mitigate, accept
Monitor: Regularly check risk status

Examples

Case 1: Feature Development Planning (Complete Tool Call Process)

User requirement: Implement search functionality

Complete Execution Process:

Step 1: Read prompt-enhancer artifact

# 使用 Bash 检测最新工件
ls -t memory/artifacts/prompt-enhancer/*.json | grep -v latest.json | head -1
# 输出：memory/artifacts/prompt-enhancer/result-20260325-103000.json

# 使用 Read 工具读取工件
Read(
  file_path="memory/artifacts/prompt-enhancer/result-20260325-103000.json"
)

# 提取需求信息
{
  "skill": "prompt-enhancer",
  "version": "2.0.0",
  "requirement": {
    "original": "实现搜索功能",
    "refined": {
      "type": "功能开发",
      "core_features": ["关键词搜索", "结果排序", "搜索历史"],
      "target_users": "普通用户",
      "platform": "小程序",
      "tech_stack": ["云开发", "云函数"],
      "success_criteria": ["搜索响应<2秒", "结果准确率>90%"],
      "timeline": "1周"
    }
  },
  "solution": {
    "selected": "方案2：云开发数据库搜索",
    "reasons": ["成本低", "维护简单", "数据量适中"],
    "alternatives": [
      {
        "name": "方案1：数据库LIKE查询",
        "pros": ["实现简单"],
        "cons": ["性能差，适合<1万条"]
      },
      {
        "name": "方案3：Algolia托管服务",
        "pros": ["功能强大", "响应快"],
        "cons": ["成本高", "需要集成"]
      }
    ]
  }
}

Step 2: Step decomposition (MECE principle)

Decompose into executable implementation steps:

# 实施计划

## 基本信息
- **项目名称**: 搜索功能
- **目标**: 实现关键词搜索、结果排序、搜索历史
- **时间线**: 1周（5个工作日）
- **团队**: 1人（全栈）

---

## 实施步骤

### Phase 1: 需求确认（Day 1）

#### 步骤1: 需求确认与技术选型
**目标**: 确认搜索范围和技术方案

**任务**:
- [ ] 确认搜索范围（标题、内容、标签）
- [ ] 确认搜索方式（模糊、精确）
- [ ] 确认数据量（预计多少条记录）
- [ ] 最终确认技术方案（云开发数据库搜索）

**产出**:
- 需求确认文档
- 技术选型确认

**依赖**: 无

**风险**: 数据量未知（应对：先调研现有数据量）

---

### Phase 2: 后端开发（Day 2-3）

#### 步骤2: 后端API开发
**目标**: 实现搜索API

**任务**:
- [ ] 开发云函数：search
- [ ] 实现搜索逻辑（模糊匹配、排序）
- [ ] 实现搜索历史记录
- [ ] 测试搜索API

**产出**:
- search 云函数
- API文档

**依赖**: 步骤1

**风险**: 搜索性能不达标（应对：提前测试性能）

---

### Phase 3: 前端开发（Day 3-4）

#### 步骤3: 前端UI开发
**目标**: 实现搜索界面

**任务**:
- [ ] 设计搜索框UI
- [ ] 实现搜索结果列表
- [ ] 实现加载状态和空状态
- [ ] 实现搜索历史展示
- [ ] API对接

**产出**:
- 搜索页面
- 搜索历史组件

**依赖**: 步骤2

**风险**: 无

---

### Phase 4: 测试上线（Day 5）

#### 步骤4: 测试与优化
**目标**: 全面测试，修复Bug

**任务**:
- [ ] 功能测试（搜索、排序、历史）
- [ ] 性能测试（响应时间）
- [ ] 边界测试（空搜索、特殊字符）
- [ ] Bug修复
- [ ] 上线准备

**产出**:
- 测试报告
- 上线代码

**依赖**: 步骤3

**风险**: 发现重大Bug（应对：预留buffer）

---

## 依赖关系图

Step 1: Requirement Confirmation ↓ Step 2: Backend Development ↓ Step 3: Frontend Development ↓ Step 4: Testing and Launch


---

## 资源规划

| 步骤 | 人员 | 时间 | 关键技术 |
|------|------|------|---------|
| 需求确认 | 1人 | 1天 | - |
| 后端开发 | 1人 | 2天 | 云开发、云函数 |
| 前端开发 | 1人 | 2天 | 小程序UI |
| 测试上线 | 1人 | 1天 | - |

**总计**: 1人 × 5天

---

## 风险与应对

| 风险 | 概率 | 影响 | 应对措施 |
|------|------|------|---------|
| 数据量未知 | 高 | 中 | 先调研现有数据量 |
| 搜索性能不达标 | 中 | 高 | 提前测试性能 |
| 发现重大Bug | 低 | 高 | 预留buffer |

---

## 关键里程碑

- **Day 1**: 需求确认完成
- **Day 3**: 后端API完成
- **Day 4**: 前端UI完成
- **Day 5**: 测试通过，准备上线

---

## 成功标准

- ✅ 搜索响应时间 < 2秒
- ✅ 搜索结果准确率 > 90%
- ✅ 搜索历史功能正常
- ✅ 无P0/P1级Bug

Step 3: plan-review (using AskUserQuestion)

AskUserQuestion(
  questions=[
    {
      "question": "实施计划已完成，请确认是否开始执行？",
      "header": "计划确认",
      "options": [
        {
          "label": "✅ 确认，开始执行",
          "description": "同意实施计划，生成工件并传递给后续技能"
        },
        {
          "label": "需要调整",
          "description": "对部分步骤有疑问，需要修改计划"
        },
        {
          "label": "重新规划",
          "description": "对整体方案不满意，需要重新分解"
        }
      ],
      "multiSelect": false
    }
  ]
)

# 用户选择：✅ 确认，开始执行

Step 4: Generate artifact (using Write)

Write(
  file_path="memory/artifacts/planning/result-20260325-104500.json",
  content={
    "skill": "planning",
    "version": "2.0.0",
    "timestamp": "2026-03-25T10:45:00Z",
    "project": {
      "name": "搜索功能",
      "goal": "实现关键词搜索、结果排序、搜索历史",
      "timeline": "1周（5个工作日）",
      "team": "1人（全栈）"
    },
    "implementation_plan": {
      "phases": [
        {
          "name": "Phase 1: 需求确认",
          "steps": [
            {
              "id": 1,
              "name": "需求确认与技术选型",
              "duration": "1天",
              "tasks": ["确认搜索范围", "确认搜索方式", "确认数据量", "确认技术方案"],
              "dependencies": [],
              "risks": ["数据量未知"]
            }
          ]
        },
        {
          "name": "Phase 2: 后端开发",
          "steps": [
            {
              "id": 2,
              "name": "后端API开发",
              "duration": "2天",
              "tasks": ["开发search云函数", "实现搜索逻辑", "实现搜索历史", "测试API"],
              "dependencies": [1],
              "risks": ["搜索性能不达标"]
            }
          ]
        },
        {
          "name": "Phase 3: 前端开发",
          "steps": [
            {
              "id": 3,
              "name": "前端UI开发",
              "duration": "2天",
              "tasks": ["设计搜索框UI", "实现搜索结果列表", "实现搜索历史展示", "API对接"],
              "dependencies": [2],
              "risks": []
            }
          ]
        },
        {
          "name": "Phase 4: 测试上线",
          "steps": [
            {
              "id": 4,
              "name": "测试与优化",
              "duration": "1天",
              "tasks": ["功能测试", "性能测试", "边界测试", "Bug修复", "上线准备"],
              "dependencies": [3],
              "risks": ["发现重大Bug"]
            }
          ]
        }
      ],
      "dependencies_graph": {
        "1": [],
        "2": [1],
        "3": [2],
        "4": [3]
      },
      "total_steps": 4,
      "total_duration": "5天"
    },
    "resource_plan": {
      "human": {
        "team_size": 1,
        "roles": ["全栈开发"]
      },
      "time": {
        "total_days": 5,
        "start_date": "2026-03-25",
        "end_date": "2026-03-29"
      },
      "technology": ["小程序", "云开发", "云函数"],
      "cost": {
        "云开发": "免费额度"
      }
    },
    "risk_assessment": [
      {
        "risk": "数据量未知",
        "probability": "high",
        "impact": "medium",
        "mitigation": "先调研现有数据量"
      },
      {
        "risk": "搜索性能不达标",
        "probability": "medium",
        "impact": "high",
        "mitigation": "提前测试性能"
      },
      {
        "risk": "发现重大Bug",
        "probability": "low",
        "impact": "high",
        "mitigation": "预留buffer"
      }
    ],
    "milestones": [
      {"day": 1, "name": "需求确认完成"},
      {"day": 3, "name": "后端API完成"},
      {"day": 4, "name": "前端UI完成"},
      {"day": 5, "name": "测试通过，准备上线"}
    ],
    "success_criteria": [
      "搜索响应时间 < 2秒",
      "搜索结果准确率 > 90%",
      "搜索历史功能正常",
      "无P0/P1级Bug"
    ],
    "review_status": "confirmed",
    "next_skills": ["experience-manager"]
  }
)

# 创建 latest.json 符号链接
ln -sf memory/artifacts/planning/result-20260325-104500.json memory/artifacts/planning/latest.json

Result: The complete tool invocation process, from reading artifacts to generating artifacts, demonstrating the actual execution process of the planning skill.

Case 2: Architecture Refactoring Planning

Requirement: Implement search functionality

Planning Result：

## 实施计划

### 步骤分解
1. 需求确认（0.5天）
   - 确认搜索范围（标题、内容、标签？）
   - 确认搜索方式（模糊、精确？）

2. 技术选型（0.5天）
   - 方案1：数据库LIKE查询（适合<1万条）
   - 方案2：Elasticsearch（适合>10万条）
   - 方案3：Algolia（托管服务，快速集成）

3. 后端开发（2天）
   - 搭建搜索API
   - 实现搜索逻辑
   - 性能优化

4. 前端开发（2天）
   - 搜索框UI
   - 结果展示
   - 交互优化

5. 测试上线（1天）
   - 功能测试
   - 性能测试
   - Bug修复

**总时长**: 6天
**风险**: 数据量未知（应对：先调研数据量）

Case 2: Architecture Refactoring Planning

Requirement: Microservice decomposition

Planning Result：

## 实施计划

### Phase 1: 调研设计（Week 1）
- 服务边界划分
- API设计
- 数据库拆分方案

### Phase 2: 基础设施（Week 2）
- 容器化
- 服务注册发现
- 配置中心

### Phase 3: 核心服务拆分（Week 3-4）
- 用户服务
- 订单服务
- 商品服务

### Phase 4: 测试迁移（Week 5）
- 集成测试
- 性能测试
- 灰度迁移

**总时长**: 5周
**风险**: 数据迁移（应对：制定详细迁移方案）

Common Pitfalls

Pitfall 1: Skipping planning and executing directly

Manifestation: Starting to write code directly once the requirements are clear Correct Approach: Enforce planning and generate an implementation plan

Pitfall 2: The plan is too coarse

Manifestation: The plan has only two steps, "development" and "testing" Correct Approach: Break it down to a granularity of 0.5-2 days

Pitfall 3: Ignoring dependencies

Manifestation: Executing dependent tasks in parallel, causing rework Correct Approach: Clarify dependencies and arrange the order reasonably

Pitfall 4: Not identifying risks

Manifestation: There is no risk assessment in the plan Correct Approach: Identify potential risks for every step

Pitfall 5: Skipping plan-review

Manifestation: Executing directly after generating the plan without confirming with the user Correct Approach: Enforce plan-review and execute only after user confirmation

References

A Guide to the Project Management Body of Knowledge (PMBOK)
The Mythical Man-Month - Frederick Brooks
WBS Work Breakdown Structure
Critical Path Method (CPM)
Risk Management Framework

About this skill

Prerequisite Protocol

Mandatory Trigger Rules

Dependency Check

Collaboration with goal-oriented

Artifact Transfer Mechanism

Input Artifact (from prompt-enhancer)

Output Artifact (passed to subsequent skills)

Overview

When to Use

The Process

Step 1: Receive Requirement Solution Artifact

Step 2: Step Breakdown

Breakdown Principles

Decomposition methods

Step 3: Dependency identification

Dependency types

Dependency graph example

Step 4: Resource planning

Resource dimensions

Resource planning table

Step 5: Risk assessment

Risk types

Risk Matrix

Step 6: Generate Implementation Plan

Implementation Plan Template

Step 7: plan-review (mandatory)

plan-review process

plan-review tool call (AskUserQuestion)

Step 8: Generate Artifact

Planning Methodology

WBS Work Breakdown Structure

Critical Path Method (CPM)

Risk Management Framework

Examples

Case 1: Feature Development Planning (Complete Tool Call Process)

Step 1: Read prompt-enhancer artifact

Step 2: Step decomposition (MECE principle)

Step 3: plan-review (using AskUserQuestion)

Step 4: Generate artifact (using Write)

Case 2: Architecture Refactoring Planning

Case 2: Architecture Refactoring Planning

Common Pitfalls

Pitfall 1: Skipping planning and executing directly

Pitfall 2: The plan is too coarse

Pitfall 3: Ignoring dependencies

Pitfall 4: Not identifying risks

Pitfall 5: Skipping plan-review

References